ChrisLomont

Changing a prior after the fact is the opposite of reasonable.

It's called a prior for a reason.

Which economic system do you think has proven better at raising standards of living?

You can pull it up. It doesn't change that it was an empty political stunt, that most guns were not in the buyback, that people simply took the buyback money to buy new guns (which you can check at the time by massive gun profits selling guns to Australians), or, as I pointed out above, that there are more guns now than before Port Authur [1].

So go ahead and post the video. Then read the rest of the story.

[1] https://australiainstitute.org.au/post/australia-more-guns-n...

120Gbps repeaters will be expensive. Making circuits that fast is incredibly tricky, with power, heat, rf (and more) issues.

If you think it's cheap make one and get rich.

But it's most certainly not cheap or easy or low cost.

They did no such thing. Australians own more guns than ever. They had a buyback program, and people used it to turn old guns into new ones.

>For each training example gradients are computed, then averaged per batch

No, that would be incredibly slow. The loss is summed over items in a batch, then the gradient is computed once based on the loss, then backpropagation is done. It's done this way by default both in TensorFlow and in Pytorch.

It is possible to sum per entry (see here [1] in pytorch docs) but is is extremely rare. The majority of training loops (this one included) is the standard "do a batch" then call "loss.backward()" which computes only once per batch.

Look at the "backward()" call in pytorch docs [2]: This "Computes the gradient of current tensor w.r.t. graph leaves." If this were done per item and being summed, there would be no need for this in the training loop, and nearly every training loop I've seen has this.

Here's pytorch explaining all this in detail [3].

And in any case, since the functions in the pipeline contain many non-smooth terms, the information required to restore an image, even if only one were done, is not in the gradient, any more than the slope of a line tells you the x-intercept.

Another way to think of it: even for perfectly smooth functions, derivatives lose information that cannot be recovered.

[1] https://pytorch.org/functorch/stable/notebooks/per_sample_gr...

[2] https://pytorch.org/docs/stable/generated/torch.Tensor.backw...

[3] https://pytorch.org/tutorials/beginner/blitz/autograd_tutori...

>No, again. The AI can be human operated, and thus it can produce copyrightable works.

And which work in the entire thread up to you moving those goalposts was not human operated?

Blog post: "Stable Diffusion is a new “text-to-image diffusion model”". "you can try..." "type in a text prompt..." "and added a prompt..." and on and on.

In every example in the post, the images were created by human inputs.

Top post in this chain: "Are the AI generated images copyrighted?" Notice the "the" instead of "Are AI generated images copyrighted?" Did you miss that word? What "the AI" do you think this refers to? Some other AI that you made up to split hairs over, or "the" AI under discussion that is creating the art in this thread?

Did you notice no one except you in the replies to that comment misunderstood it? Not one person misread it, except you.

>it's you not understanding what the premise is.

Not a single item in the chain above you was about anything other than humans using AI to make images. You made up a premise to split hairs over, but it looks like that is your modus operandi judging from past posts.

I think you need to read the thread carefully before claiming others don't understand.

Have a good day.

This AI, in this thread, is human operated. Those images are likely copyrighted.

What case are you using to claim AI images are not copyrighted? The one I cited? Or another? The one I cited, the only one I can find, is not relevant here.

Why are you avoiding citing a case to back up your claim?

To

>> Are the AI generated images copyrighted?

you wrote

> No, this is already established in the US.

What cases are you then using to conclude this?

>It is not copyrighted by default anymore than any other trivial non de minimis effort

I doubt de minimus applies, since the images coming out of this AI are no where near any originals that I've seen - they may similar style, but they're so completely different that they are almost all certainly copyrighted by the author.

For example, the simplest, littlest effort to snap a picture gives it full copyright. This AI takes more work to use than snapping a picture.

And de minimis would be on the infringer to prove - the owner still has a copyright until proven otherwise. So the images are most likely copyrighted.

Anything copyrightable in the US is automatically copyrighted by default upon creation. It's why any photo taken is immediately copyrighted by the photographer. Any painting is automatically copyrighted by the painter. Any computer art is automatically copyrighted by the computer operator.

I'd expect most of these works are copyrighted just as much as if someone painted them, or used blender to make them, or used a collage tool to merge self owned photos.

Technically, it is no where near that. I'm not even sure what a "full set of gradients" for an image would even mean - it's not a term.

Weights are usually randomly initialized, defining a function. Many images (which are randomly batched on each epoch) are passed through, and a loss function tallied. Backprop gets a gradient, and a training schedule tweaks the weight a little bit. No image has its "gradients" somehow added. The overall function has gradients, but they are not smooth, due to many things such as ReLU being non-smooth. These are not invertible in any sense.

Th networks have many, many irreversible steps in them that truncate, that max pool, that perform dropout during steps, many of these choices done randomly. There are gradient normalizations to clip or stretch to deal with vanishing or exploding gradients, and SOOO many lossy things done that are completely irreversible.

So none of this can possibly work. A network is not simply a sum of images in some format.

>Just because the image is publicly available doesn't mean it should have been copied.

It wasn't copied - it was viewed, which is perfectly legal.

The AI does not have a copy of any of the the images it learned from. It merely learned styles and concepts of many images with associated text for those styles and concepts.

>> Are the AI generated images copyrighted?

> No, this is already established in the US.

The case you're likely thinking of [1] is not at all this case. Thaler tried to have the AI assigned the copyright, then transfer it to himself. USPTO denied that the AI could have the copyright, so it could not be transferred.

This is completely the opposite of what this case is. He wanted AI to be the sole copyright holder, which the court ruled cannot be. Any creative human input to make the AI create the image (text prompts, post selection from among any images to find one that is good) is fully copyrightable by the person.

The text that an artist uses and then selects final images from are both considered creative enough to make the final copyrightable, just as if they entered numbers into Photoshop tools to make an image the way they wanted it. Plenty of these are creative enough and take significant tweaking by the authors via the text they fiddle with to be copyrightable, the same as the text they created to generate the work.

[1] https://www.copyright.gov/rulings-filings/review-board/docs/...

>At some point some AI is going to spit something out too close to something else and the courts will probably have to decide.

People have been spitting out stuff too close to decide and letting courts decide for well over a hundred years. This is nothing new.

If people make art that is in the style of someone else, by hand or by AI, it's generally fine to do so. If it gets too close, then courts can step in as always.

>Second, the trend is downward.

Yeah, there is a future that is not represented, so there is no reason to assume any downward trend. Plus he only sampled things for which there were open source alternatives (which are also questionable: e.g., AutoCAD - a full featured CAD tool versus Open Cascade, a C++ lib with no UI? Really?!).

The proper way to do such an analysis if different on almost every choice the author made:

1. Randomly, uniformly sample products/service over all time (somewhat difficult). Or better yet, make a cutoff date far enough the past to hide the fake triangle due to items being release last year.

2. Find OS alternative, it if exists, and most importantly, decide on some level of how much is is actually an alternative. Almost nothing on this list is an actual alternative, any more than a text editor is a MS Word replacement.

3. Analyze more carefully.

Post selecting products that fit a desired outcome, ignoring lots and lots of other products, ignoring the time bias, and calling things replacements that are not, then drawing these conclusions, is really bad analysis.

>it still takes years until it can outright displace the proprietary champion

How many items on even this biased selection list displaced the proprietary champion by killing their market? None? A few proprietary products died well on their own. And in this time frame, how many open source projects died?

We have. Look at the Wikipedia page on history of recessions as a place that makes it clear.

If you're implying that blacks incarcerated at 5x the rate of population is evidence of racism, then is men being incarcerated at 10x their population rate evidence of sexism?

Or is there other factors?

Ok, but the point is the same - a huge amount of human records and skills are scored against a moving backdrop of human scores that are current competitors.

Federer getting grand slams - all against local in time competitors - is impressive. Had he played 50 years ago, he'd have likely gotten more. Had he played in 50 years, likely less, expecting competition to improve.

Same for NBA records - all against a moving, relative backdrop.

Same for wrestling, baseball, hockey, and on and on and on.

Just because your score is relative to others does not make reaching some milestone, especially this one, less impressive.

Also, it's easier to have a higher ELO if others do too, not the other way around.

One person reaching 2900 when it's not by changing the scale shows their relative position to others is a new record. it's a worthy psychological goal, like breaking a 4 minute mile even though a minute and a mile are made up quantities. Or scoring 300 in bowling, when 10 rounds is arbitrary.

Pretty much every record in every sport is arbitrary, yet people like to set goals for motivation.

>You still have to decide what you take a pic of, that's the artistic part.

You still have to select the 2 AI jpegs out of the 15, that's the artistic part.

There's tons of them. Google patent lawsuit insurance.

What makes you think this? Did you look it up? Have a source?

Every 401k or other investment place I've seen allows either taking dividends or reinvesting them. I've never one seen a fund that picks only divide free or divided providing stocks as a choice.

And every big fund also provides all sorts of general index funds, which certainly contain both. And individual stocks themselves decide to pay dividends or not, and routinely change so.

I don't think your distinction is at all a relevant factor in these funds to nearly any end buyer.

Please provide a citation as to why you think your claim is true.

Profit and growth go together. In Amazon's case lots of profit is what fueled their growth: at each point they made far more money than needed to sustain, so they bought growth instead of paying out to shareholders.

The two are intricately related.

The majority of public shares are held in long term investments like 401ks, whose owners are most certainly interested in long term growth. Buy and hold is by far the biggest amount of investment type of stockholder. Most people don't care to fiddle with microtrading.

As the buffet bet made clear, and as the majority of advice and investors do, long term ownership outgrows those chasing short term returns, despite pop belief.

I think you vastly overestimate the dollar amounts behind stock owners.

Steam still has shareholders to answer to, with the same human motives as any shareholder, so I don't think this argument holds water. I suspect they'd try to get whatever growth they can because otherwise eventually someone will overtake them.

And empirically they are still growing at rates [1] not common among any class of companies, so someone there is certainty pushing for incredible growth.

[1] https://www.statista.com/statistics/308330/number-stream-use...

Isn't it possible that demographics have changed enough that funding requirements have changed enough that the 1940s-2000 or so system is no longer viable?

I know the demographics in the US has vastly changed some cost models over that time, making some earlier social program hit major funding problems through no evil from govt actors.

Plenty of languages have this in their standard libraries for exactly this purpose. I use it in C/C++ and C# frequently in numerical code.

C# supports it. CUDA supports it. ARM and related C/C++ compilers support it. Intel is adding hardware support in upcoming chips, so expect a lot more languages to add it.

float16 and float128 (as well as longer formats) were standardized in IEEE754 in 2008. Half is now built into C#, many other languages, and gaining ground in hardware support.

FDE can be done on the SSD after compression, and often is.

Adding life to SSDs is a terribly useful feature

Plenty of SSDs implement strong encryption, and there's an entire standard API for it. Why waste cpu on tasks your hardware already does?

Bitlocker supports on drive hardware encryption, and I'd be surprised if other major file systems didn't.

If I recall, it's a FIPS requirement for data at rest now.